Physical Sciences and Mathematics Commons^™

Variable Selection In Regression Using Multilayer Feedforward Network, Tejaswi S. Kamble, Dattatraya N. Kashid

Journal of Modern Applied Statistical Methods

The selection of relevant variables in the model is one of the important problems in regression analysis. Recently, a few methods were developed based on a model free approach. A multilayer feedforward neural network model was proposed for developing variable selection in regression. A simulation study and real data were used for evaluating the performance of proposed method in the presence of outliers, and multicollinearity.

The Goldilocks Dilemma: Impacts Of Multicollinearity -- A Comparison Of Simple Linear Regression, Multiple Regression, And Ordered Variable Regression Models, Grayson L. Baird, Stephen L. Bieber

Journal of Modern Applied Statistical Methods

A common consideration concerning the application of multiple linear regression is the lack of independence among predictors (multicollinearity). The main purpose of this article is to introduce an alternative method of regression originally outlined by Woolf (1951), which completely eliminates the relatedness between the predictors in a multiple predictor setting.

Symmetric Variants Of Logistic Smooth Transition Autoregressive Models: Monte Carlo Evidences, Olaoluwa S. Yaya, Olanrewaju I. Shittu

Journal of Modern Applied Statistical Methods

The Smooth Transition Autoregressive (STAR) models are becoming popular in modeling economic and financial time series. The asymmetric type of the model is the Logistic STAR (LSTAR) model, which is limited in its applications as a result of its asymmetric property, which makes it suitable for modelling specific macroeconomic time series. This study was designed to develop the Absolute Error LSTAR (AELSTAR) and Quadratic LSTAR (QLSTAR) models for improving symmetry and performance in terms of model fitness. Modified Teräsvirta’s Procedure (TP) and Escribano and Jordá's Procedure (EJP) were used to test for nonlinearity in the series. The performance of the …

Liu-Type Logistic Estimators With Optimal Shrinkage Parameter, Yasin Asar

Journal of Modern Applied Statistical Methods

Multicollinearity in logistic regression affects the variance of the maximum likelihood estimator negatively. In this study, Liu-type estimators are used to reduce the variance and overcome the multicollinearity by applying some existing ridge regression estimators to the case of logistic regression model. A Monte Carlo simulation is given to evaluate the performances of these estimators when the optimal shrinkage parameter is used in the Liu-type estimators, along with an application of real case data.

Solution To The Multicollinearity Problem By Adding Some Constant To The Diagonal, Hanan Duzan, Nurul Sima Binti Mohamaed Shariff

Journal of Modern Applied Statistical Methods

Ridge regression is an alternative to ordinary least-squares (OLS) regression. It is believed to be superior to least-squares regression in the presence of multicollinearity. The robustness of this method is investigated and comparison is made with the least squares method through simulation studies. Our results show that the system stabilizes in a region of k, where k is a positive quantity less than one and whose values depend on the degree of correlation between the independent variables. The results also illustrate that k is a linear function of the correlation between the independent variables.

Jmasm39: Algorithm For Combining Robust And Bootstrap In Multiple Linear Model Regression (Sas), Wan Muhamad Amir, Mohamad Shafiq, Hanafi A.Rahim, Puspa Liza, Azlida Aleng, Zailani Abdullah

Journal of Modern Applied Statistical Methods

The aim of bootstrapping is to approximate the sampling distribution of some estimator. An algorithm for combining method is given in SAS, along with applications and visualizations.

Jmasm35: A Percentile-Based Power Method: Simulating Multivariate Non-Normal Continuous Distributions (Sas), Jennifer Koran, Todd C. Headrick

Journal of Modern Applied Statistical Methods

The conventional power method transformation is a moment-matching technique that simulates non-normal distributions with controlled measures of skew and kurtosis. The percentile-based power method is an alternative that uses the percentiles of a distribution in lieu of moments. This article presents a SAS/IML macro that implements the percentile-based power method.

Factorial Invariance Testing Under Different Levels Of Partial Loading Invariance Within A Multiple Group Confirmatory Factor Analysis Model, Brian F. French, Holmes Finch

Journal of Modern Applied Statistical Methods

Scalar invariance in factor models is important for comparing latent means. Little work has focused on invariance testing for other model parameters under various conditions. This simulation study assesses how partial factorial invariance influences invariance testing for model parameters. Type I error inflation and parameter bias were observed.

A Comparison Of Estimation Methods For Nonlinear Mixed-Effects Models Under Model Misspecification And Data Sparseness: A Simulation Study, Jeffrey R. Harring, Junhui Liu

Journal of Modern Applied Statistical Methods

A Monte Carlo simulation is employed to investigate the performance of five estimation methods of nonlinear mixed effects models in terms of parameter recovery and efficiency of both regression coefficients and variance/covariance parameters under varying levels of data sparseness and model misspecification.

A Spatial Analytical Framework For Examining Road Traffic Crashes, Grace O. Korter

Journal of Modern Applied Statistical Methods

A number of different modeling techniques have been used to examine road traffic crashes for analytic and predictive purposes. Map-based spatial analysis is introduced. Applications are given which show the power in a combination of existing exploratory and statistical methods.

Jmasm36: Nine Pseudo R^2 Indices For Binary Logistic Regression Models (Spss), David A. Walker, Thomas J. Smith

Journal of Modern Applied Statistical Methods

This syntax program is an applied complement to Veall and Zimmermann (1994), Menard (2000), and Smith and McKenna (2013) and produces nine pseudo R² indices, not readily accessible in statistical software such as SPSS, which are used to describe the results from binary logistic regression analyses.

Jmasm38: Confidence Intervals For Kendall's Tau With Small Samples (Spss), David A. Walker

Journal of Modern Applied Statistical Methods

A syntax program, not readily expedient in statistical software such as SPSS, is provided for an application of confidence interval estimates with Kendall’s tau-b for small samples.

Almost Unbiased Estimator Using Known Value Of Population Parameter(S) In Sample Surveys, Rajesh Singh, S.B. Gupta, Sachin Malik

Journal of Modern Applied Statistical Methods

An almost unbiased estimator using known value of some population parameter(s) is proposed. A class of estimators is defined which includes Singh and Solanki (2012) and Sahai and Ray (1980), Sisodiya and Dwivedi (1981), Singh, Cauhan, Sawan, and Smarandache (2007), Upadhyaya and Singh (1984), Singh and Tailor (2003) estimators. Under simple random sampling without replacement (SRSWOR) scheme the expressions for bias and mean square error (MSE) are derived. Numerical illustrations are given.

Model-Based Outlier Detection System With Statistical Preprocessing, D. Asir Antony Gnana Singh, E. Jebalamar Leavline

Journal of Modern Applied Statistical Methods

Reliability, lack of error, and security are important improvements to quality of service. Outlier detection is a process of detecting the erroneous parts or abnormal objects in defined populations, and can contribute to secured and error-free services. Outlier detection approaches can be categorized into four types: statistic-based, unsupervised, supervised, and semi-supervised. A model-based outlier detection system with statistical preprocessing is proposed, taking advantage of the statistical approach to preprocess training data and using unsupervised learning to construct the model. The robustness of the proposed system is evaluated using the performance evaluation metrics sum of squared error (SSE) and time to …

An Evaluation Of Pareto, Lognormal And Pps Distributions: The Size Distribution Of Cities In Kerala, India, Christopher A. Vallabados, Subbarayan A. Arumugam

Journal of Modern Applied Statistical Methods

The Pareto-Positive Stable (PPS) distribution is introduced as a new model for describing city size data of a region in a country. The PPS distribution provides a flexible model for fitting the entire range of a set of city size data and the classical Pareto and Zipf distributions are included as a particular case.

The Xgamma Distribution: Statistical Properties And Application, Subhradev Sen, Sudhansu S. Maiti, N. Chandra

Journal of Modern Applied Statistical Methods

A new probability distribution, the xgamma distribution, is proposed and studied. The distribution is generated as a special finite mixture of exponential and gamma distributions and hence the name proposed. Various mathematical, structural, and survival properties of the xgamma distribution are derived, and it is found that in many cases the xgamma has more flexibility than the exponential distribution. To evaluate the comparative behavior, stochastic ordering of the distribution is studied. To estimate the model parameter, the method of moment and the method of maximum likelihood estimation are proposed. A simulation algorithm to generate random samples from the xgamma distribution …

Analyzing Different Sampling Designs (Sas), Ying Lu

Journal of Modern Applied Statistical Methods

Various sampling designs are reviewed within the framework of probability sampling. SAS® code to estimate means and proportions, and their standard errors, using different sampling designs are illustrated using example data sets.

Determination Of Optimal Tightened Normal Tightened Plan Using A Genetic Algorithm, Sampath Sundaram, Deepa S. Parthasarathy

Journal of Modern Applied Statistical Methods

Designing a tightened normal tightened sampling plan requires sample sizes and acceptance number with switching criterion. An evolutionary algorithm, the genetic algorithm, is designed to identify optimal sample sizes and acceptance number of a tightened normal tightened sampling plan for a specified consumer’s risk, producer’s risk, and switching criterion. Optimal sample sizes and acceptance number are obtained by implementing the genetic algorithm. Tables are reported for various choices of switching criterion, consumer’s quality level, and producer’s quality level.

Bayesian Estimation Of P[Y < X] Based On Record Values From The Lomax Distribution And Mcmc Technique, Mohamed A. W Mahmoud, Rashad M. El-Sagheer, Ahmed A. Soliman, Ahmed H. Abd Ellah

Journal of Modern Applied Statistical Methods

Our interest is in estimating the stress-strength reliability R = P[Y < X], where X and Y follow the Lomax distribution with common scale parameter. We discuss the problem in the situation where the stress measurements and the strength measurements are both in terms of records. Firstly, we obtain the MLE of R in general case (the common scale parameter is unknown). The MLE of the three unknown parameters can be obtained by solving one non-linear equation. We provide a simple fixed point type algorithm to find the MLE. We propose percentile bootstrap confidence intervals of R. A Bayes …

Generalized Linear Model Analyses For Treatment Group Equality When Data Are Non-Normal, Harvey J. Kesleman, Abdul R. Othman, Rand R. Wilcox

Journal of Modern Applied Statistical Methods

One of the validity conditions of classical test statistics (e.g., Student’s t-test, the ANOVA and MANOVA F-tests) is that data be normally distributed in the populations. When this and/or other derivational assumptions do not hold the classical test statistic can be prone to too many Type I errors (i.e., falsely rejecting too often) and/or have low power (i.e., failing to reject when the null hypothesis is false) to detect treatment effects when they are present. However, alternative procedures are available for assessing equality of treatment group effects when data are non-normal. For example, researchers can use robust estimators …

Construction Of Pair-Wise Balanced Design, Rajarathinam Arunachalam, Mahalakshmi Sivasubramanian, Dilip Kumar Ghosh

Journal of Modern Applied Statistical Methods

A new procedure for construction of pair wise balanced design with equal replication and un-equal block sizes based on factorial design have been evolved. Numerical illustration also provided. It was found that the constructed pair wise balanced design was found to be universal optimal.

Flesch-Kincaid Reading Grade Level Re-Examined: Creating A Uniform Method For Calculating Readability On A Certification Exam, Emily Neuhoff, Kristiana M. Feeser, Kayla Sutherland, Thomas Hovatter

Online Journal for Workforce Education and Development

Abstract

Objective: This study attempted to establish a consistent measurement technique of the readability of a state-wide Certified Nursing Assistant’s (CNA) certification exam. Background: Monitoring the readability level of an exam helps ensure all test versions do not exceed the maximum reading level of the exam, and that knowledge of the subject matter, rather than reading ability, is being assessed. Method: A two part approach was used to specify and evaluate readability. First, two methods (Microsoft Word^® (MSW) software and published readability formulae) were used to calculate Flesch Reading Ease (FRE) and Flesch-Kincaid Reading Grade Level (FKRGL) for multiple …